Journal of FinancialServices Research9:241-257 (1995) 1995 KinwerAcademic Publishers Financial Fragility: Issues and Policy Implications CHARLES W. CALOMIRIS Department of Finance, University of Illinois at Urbana-Champaign, 340 Commerce West, Champaign, IL 61820 1. Introduction This article addresses the question of how financial institutions, contracting forms, and government financial policies affect the degree of macroeconomic volatility. Models that posit such relationships are sometimes referred to as models of "financial fragility." These models explore ways in which the financial system can add to the volatility of economic activity by defining sources of financial "shocks" and financial "propagators" of other shocks. Financial shocks are defined as disturbances to the real economy that originate in financial markets. Financial propagation refers to the ways in which financial contracts, markets, and intermediaries can serve to aggravate shocks that originate else- where. Economists have not always been receptive to the idea that financial arrangements matter for business cycles. From the standpoint of traditional neoclassical general equi- librium theory, financial arrangements (which include financial contracting, the actions of financial intermediaries, and government policies toward the financial sector) typically are viewed as epiphenomenal--simply as a set of mechanisms for executing Arrow- Debreu contingent claims to allocate resources optimally. Mainstream macroeconomists and finance specialists of the 1960s seemed to agree. 1 Corporation financial decisions were neutral according to Modigliani and Miller (1958), with the addition of minor caveats to take account of physical bankruptcy costs and tax incentives; and firms all faced identical costs of funds adjusted for systematic risk factors according to the capital asset pricing model. Thus, there was no call to object to the standard IS-LM macroeconomic framework's assumption that all firms effectively faced the same cost of funds (summa- rized by "the" interest rate) and that this cost equaled the marginal product of capital. One of the reasons that macroeconomic theorists and financial economists have changed paradigms in recent years, and have come to view financial relationships as more than epiphenomenal, is the growing evidence that financial relationships affect the volatility of economic activity. Such effects fall into two categories--namely, financial shocks to the macroeconomy, and financial propagators of disturbances that originate elsewhere. 2 Models that incorporate these financial effects imply "excessive" volatility of output from the standpoint of the neoclassical benchmark model. In an environment that allows financial arrangements to act as sources and propagators of shocks, output will vary more than in an environment in which financial arrangements are merely epiphenomenal. 47 242 CHARLESW. CALOMIRIS In this article, I will briefly review a small subset of the theory and evidence that underlie the argument that financial factors can contribute to macroeconomic volatility. I do not provide a comprehensive review of the literature here; rather my aim is to illustrate impor- tant categories of effects and supporting evidence for those ideas. 3 In conclusion, I consider appropriate policy responses in light of theory and evidence. 2. Financial propagators Some of the most important financial propagators include: (1) the cash-flow constraint, (2) balance-sheet (or more narrowly, leverage) constraints, (3) external supply-of-funds constraints (limitations on bank credit), and (4) financial regulations that magnify busi- ness cycles. These constraints really are not different phenomena, but different aspects of the same underlying costs of information and control attendant to corporate finance. Each acts as a financial propagator, in the sense that, relative to the benchmark of neoclassical financial neutrality, the economic activity of firms is more volatile in response to variation in exogenous disturbances as a result of these related factors. 2.1. The cash-flow constraint The "cash-flow constraint" is simply shorthand for the statement that firms raise funds more cheaply internally than externally. Information and transaction costs account for the wedge between the cost of internal and external funds (e.g., Stiglitz and Weiss, 1981; Bernanke and Gertler, 1990; Greenwald and Stiglitz, 1990; Calomiris and Hubbard, 1990). This constraint increases the sensitivity of investment to changes in firm earnings (a "financial accelerator"), and thus naakes investment more volatile that under the standard neoclassical, "flexible-accelerator" model. Empirical evidence from panel data studies emphasizes that this effect is not uniform across firms (Fazzari, Hubbard, and Petersen, 1988; Calomiris and Hubbard, 1995; Bernanke, Gertler, and Gilchrist, 1994; Calomiris, Himmelberg, and Wachtel, 1995; Kashyap, Owen, and Stein, 1993; Carpenter, Fazzari, and Petersen, 1993; Himmelberg and Petersen, 1993). Firms with identifiably higher costs of external finance display much greater cash-flow sensitivity of fixed invest- ment and inventory accumulation, even after controlling for fundamental investment opportunities (using innovations in sales, or measures of Tobin's Q).4 Calomiris and Hubbard (1995) analyze the characteristics and behavior of different classes of firms, divided according to their shadow costs of external finance. The shadow cost of external finance is derived from the firms' responses to the undistributed profits tax of 1936, which placed a progressive surtax on retained earnings: They find that firms with high external finance costs tend to be small, tend to have high debt ratios and low ratios of profits to sales, and tend to experience high profits growth. Calomiris and 48 FINANCIAL FRAGILITY: ISSUES AND POLICY IMPLICATIONS 243 Hubbard argue that these attributes can be seen as characteristics of "unseasoned" credit risks, that is, of fast growing firms, that require the discipline of short-term debt when financing themselves externally, and which have not achieved their long-run scale of operations. Table 1 shows that firms facing high costs of external finance exhibit much greater cash-flow sensitMty of fixed investment and working capital investment than other firms, Table 1. The cash-flow sensitivity of investment in fixed capital and working capital by finns' costs of external finance (A = Low, B = Middle, C = High) Fixed Capital Investment Regressions, 1936 Regression number (I) (2) Dependent variable I1936/K1935 I1936/K1935 Number of observations 244 244 Adjusted R-squared 0.063 0.217 Standard Standard Coefficient error Coefficient error Constant - 0.019 0.022 0.015 0.021 Type B - 0.037 0.036 Type C - 0.1t2 0.051 Q1935 0.044 0.016 0.024 0.011 Q1935 x Type B 0.024 0.019 Q1935 x Type C 0.039 0.051 (CF1935 q- CF1936)fKI935 0.018 0.016 -0.004 0.014 [(CF1935 + CF1936)/K1935] x Type B 0.003 0.018 [(CF1935 + CF1936)/K1935] x Type C 0.248 0.100 Change in Working Capital Regression, 1936 Regression number (1) (2) Dependent variable AWK1936/K1935 _~WK1936/KI935 Number of observations 244 244 Adjusted R-squared 0.276 0.209 Standard Standard Coefficient error Coefficient error Constant - 0.016 0.047 - 0.017 0.048 Type B - 0.025 0.076 - 0.031 0.079 Type C - 0.211 0,385 - 0.149 0.300 Q1935 0.054 0.040 0.046 0.037 Q1935 x Type B -0.026 0.067 0.002 0.068 Q1935 x Type C 0.171 0.257 0.354 0.355 (CF1935 + CF1936)/KI935 - 0.028 0.056 [(CFI935 + CF1936)fK1935] X Type B 0.214 0.061 [(CFI935+CF1936)/K1935 ] x Type C 1.005 0.737 CF1935/K1935 - 0.023 0.093 CF1935/KI935 x Type B 0.390 0.100 CFI935fKI935 x Type C 0.841 1.447 Note: Heteroscedasticity-consistent standard errors are presented. Source: CaIomiris and Hubbard (1995). 49 244 CHARLES W. CALOMIRIS after controlling for firm opportunities as measured by Tobin's Q. Moreover, the sensi- tivity of working capital is especially pronounced, which is consistent with the notion that firms with high finance costs self-insure against cash-flow fluctuations with working capital. They accumulate liquid assets during high earnings periods and draw them down during low earnings periods. Other firms, with lower cost access to external finance need not follow this same practice. Calomiris, Himmelberg, and Wachtel (1995) study the characteristics of firms that differ in terms of access to credit markets. At the top of the "pecking order" are firms with access to commercial paper and rated bond markets; next are those with access to rated bond markets only; and at the bottom are firms without any rated debt. The researchers find that the size of firms, and their stocks of liquid assets, vary importantly across credit classes. Firms with limited access to credit markets tend to be smaller firms with larger stocks of liquid assets (inventories and financial working capital). As in Calomiris and Hubbard (1995), Calomiris, Himmelberg, and Wachtel show that the sensitivity of a firm's inventory accumulation or financial working capital investment decreases with the firm's access to sources of external finance. The conclusions of these and other recent panel studies of cash-flow sensitivity are that fixed investment, R&D investment, and working capital investment are all "excessively" sensitive to cash flow (relative to the benchmark neoclassical model of investment). Sensitivity to cash flow (or "financial slack") varies systematically with the firm's costs of external finance. Liquid assets (inventories, cash, and net